Is a Dark Virtual Environment Scary?

CYBERPSYCHOLOGY & BEHAVIOR Volume 12, Number 4, 2009 ª Mary Ann Liebert, Inc. DOI: 10.1089=cpb.2008.0293

ORIGINAL ARTICLE

Is a Dark Virtual Environment Scary? Alexander Toet, Ph.D.,1 Marloes van Welie, M.Sc.,1 and Joske Houtkamp, Ph.D.2

Abstract

This study investigated the effects of nighttime lighting conditions and stress on the affective appraisal of a virtual environment (VE). The effective application of VEs in emotionally intense simulations requires precise control over their characteristics that affect the user’s emotions and behavior. It is known that humans have an innate fear of darkness, which increases after exposure to stress and extrapolates to ecologically valid (immersive) VEs. This study investigated if the simulated level of illumination determines the affective appraisal of a VE, particularly after stress. Participants explored either a daytime or a nighttime version of a VE, after performing either an acute psychosocial stress task (Trier Social Stress Test, or TSST) or a relaxing control task. The affective qualities of the VE were appraised through the Russel and Pratt semantic questionnaire on the valence and arousal dimensions. Distress was assessed through free salivary cortisol, the state self-report scale from the Spielberger State-Trait Anxiety Inventory (STAI), and heart rate. In addition, memory for scenic details was tested through a yes–no recognition test. Free salivary cortisol levels, heart rates, and scores on the STAI all indicate that participants who were subjected to the stress task indeed showed signs of distress, whereas participants in the control group showed no signs of stress. The results of the semantic questionnaire and the recognition test showed no significant overall effect of time-of-day conditions on the affective appraisal of the VE or on the recognition of its details, even after prior stress. The experiences of users exploring the VE were not affected by the simulated lighting conditions, even after acute prior stress. Thus, lowering the illumination level in a desktop VE is not sufficient to elicit anxiety. Hence, desktop VE representations are different from immersive VE representations in this respect. This finding has implications for desktop VE representations that are deployed for serious gaming and training purposes.

Introduction

T

he present study investigated whether users appraise the nighttime representation of a virtual environment (VE) as less pleasant and more arousing than its daytime version, particularly after experiencing acute stress. Desktop VEs are increasingly deployed to train personnel to operate in dangerous or critical situations, where emotions influence performance. The effectiveness of desktop VEs for such training purposes depends critically on their ability to correctly address the user’s emotional, cognitive, and perceptual experience.1 Emotionally compelling VEs provide more effective training1,2 and result in a high degree of initial learning and subsequent retention of the lessons learned.3 Although desktop VEs have successfully been used to induce affective responses,4 the relation between their specific content and the user response remains unclear. For practical applications in emotionally intense simulations, the design and use of desktop VEs requires precise knowledge of and control over

the characteristics that determine the user’s perception and cognition and that affect his or her emotions and behavior. Darkness is a phylogenetically relevant aversive context for diurnal creatures like humans. It connotes feelings of personal vulnerability to actual physical danger and may automatically precipitate emotional responses consonant with those thoughts. Whereas ambient light may be reassuring, ambient darkness elicits fear by concealing potential dangers.5–8 Evidence for the anxiogenic nature of darkness is provided by the fact that the human startle reflex increases in the dark.9,10 The innate fear of darkness that most people have also extrapolates to ecologically valid virtual (immersive) environments.11 In commercial desktop games, low-key lighting is sometimes deployed to evoke suspense and dread.12 It has also been shown recently that color of simulated illumination in desktop VEs determines the user’s affect and performance.13 However, it is currently not known if the level of ambient lighting actually determines the affective appraisal of a desktop VE.

1

Department of Applied Perception, TNO Human Factors, Soesterberg, The Netherlands. Institute of Information and Computing Sciences, University of Utrecht, The Netherlands.

2

363

364 Affective appraisals are judgments concerning the capacity of the appraised environment to alter an individual’s mood, expressed in terms such as pleasant, repulsive, and attractive.14 The affective appraisal of an environment is important because it continually, and often unconsciously, modifies our emotions, mood, and behavior. People’s affective appraisals of physical environments vary mainly along two dimensions: arousal and pleasure.15,16 Because the affective appraisal process also has a cognitive component, a person’s internal affective state may in turn influence the perception of the affective qualities of an environment through misattribution.17–20 People are inclined to make cognitive appraisals of unrelated topics and objects reflecting their affective state (e.g., induced by watching movies, enjoying sunny weather, or experiencing stressful exams).19,21 Also, they tend to attribute residual arousal from prior situations to external cues in subsequent situations.22 Thus, people may also misattribute their own feelings to the affective qualities of VEs. Evidence for this hypothesis is our recent finding that users suffering from cybersickness experience a simulated environment as less pleasant and more arousing than users who are not afflicted.23 Anxiety generates a hypervigilance for threatening stimuli,24 induces selective attentional bias for anxiogenic cues,25 causes a focusing on cues that support the subjective feeling of threat,26 and impairs auditory perception,27 thus reducing the intake of cues and information upon which to make accurate appraisals and attributions. In addition, darkness-induced anxiety is enhanced by prior exposure to a social stressor.28 Therefore, we hypothesized that participants would appraise a nighttime version of a desktop VE as less pleasant and as more arousing than its equivalent daytime VE. To test this hypothesis, we first exposed 52 young male participants either to a nonstressful control (reading) task or to the Trier Social Stress Test (TSST),29,30 a validated psychosocial stress task. After performing this initial task, they explored either a daytime or a nighttime version of a virtual environment representing a small village with typical Italian architecture. To assess their anxiety level, we measured free salivary cortisol and heart rate, and we obtained some selfreport questionnaires. The affective appraisal of the VE was measured using the Russel and Pratt Semantic questionnaire.15 In addition, we tested a possible anxiety-induced narrowing of attention for nonthreatening environmental cues (due to a focusing on threatening stimuli) through a recognition test in which participants were presented with screenshots depicting details from the VE, both from parts they had explored during the experiment and from areas they had not encountered. Our secondary hypothesis was that participants in the nighttime condition would recognize less neutral details than participants in the daytime condition, and even more so after prior exposure to acute social stress. To the best of our knowledge, this is the first study to specifically investigate the effects of darkness and prior stress on the affective appraisal of a desktop VE. Methods Experimental environment The experiments were performed in two separate rooms. Both rooms were windowless, temperature-controlled, and

TOET ET AL. sound-attenuated, and their doors were closed during the experiments. The first room (A) contained the computer system that was used to present the VE, a CRT display and video recorder that were used to monitor and register the events in an adjacent room (B), and some additional chairs and a table. The second room (B) was equipped with a camera and contained an office table with three chairs. Heart rate measurement Heart rate was continuously monitored at a sampling frequency of 1 Hz during the entire period of the experiment, using a chest strap Polar Wearlink 31 wireless transmitter and a wrist strap Polar RS400 Running Computer (www.polar.fi). After the experiment, heart rates were averaged over 1minute intervals. Saliva sampling and biochemical analysis Cortisol, a neurohormone released when the stress system is aroused, was used to index stress responsivity. When humans perceive an event as stressful, a cascade of physiological events occurs along the hypothalamic-pituitary-adrenal (HPA) axis, which result in cortisol release.31 Cortisol reactivity in response to stress is greatest for young men.32 Salivary measures of cortisol are a valid and reliable reflection of serum cortisol.33,34 Saliva samples for the assessment of free salivary cortisol were collected from all 52 participants, both immediately before and again 35 minutes after the start of the experiment (i.e., 15 min after the start of the TSST, when cortisol levels peak29), using Salivette cortisol collection devices (Sarstedt, Germany; www.sarstedt.com). Salivettes were capped, labeled, and stored at
208C after completion of the session until biochemical analysis took place. The free cortisol concentrations in the saliva samples were subsequently determined with a competition-based luminescence immunoassay kit (IBL Inc., Hamburg, Germany) using a time-resolved immunoassay with fluorometric detection.35 The intraassay and interassay variability was less than 10%, according to the manufacturer. Translated state anxiety inventory A validated Dutch translation of the state self-report scale36 from the Spielberger State-Trait Anxiety Inventory (STAI)37 was administered to assess how anxious participants felt both at the start and at the end of the experiment. This scale, which contains 20 items that can each be rated on a 4-point Likert scale, is specifically sensitive to changes in anxiety and can therefore be used to evaluate the effect of interventions. General questionnaire A general questionnaire was designed to gather information about the participants’ personality. It includes two questions about the preferred time of working and six questions from the Big Five questionnaire,38 three measuring the level of extraversion and the other three measuring the level of neuroticism. Since extraversion39 and neuroticism40 are known to correlate positively with stress, these factors were used as covariates in the analysis of the results. The two questions about the preferred time of working were not further used in the analysis.

IS A DARK VIRTUAL ENVIRONMENT SCARY? Game Experience questionnaire A Game Experience questionnaire was designed to assess the participants’ experience with games and VEs in general and with Counter-Strike
in particular. It comprises nine questions related to the total amount of time spent on playing computer games, the experience with navigating VEs, and the kind of games played. The Trier Social Stress Test The TSST is a well-established psychosocial and cognitive task that can be deployed to induce acute stress in laboratory conditions.29 Its protocol consists of a 10-minute preparation period (causing anticipatory stress), followed by a 10-minute test period (a 5-minutes public speaking task followed by a demanding 5-minute oral arithmetic task: e.g., serially counting down from 1,022 in increments of 13), both performed standing before a critical jury.29,30 The TSST reliably provokes significant changes of cardiovascular parameters, subjective distress ratings, and an increase of cortisol levels,29 even in a virtual reality setting.41 At the start of the TSST, the investigator led the participant from room A to room B, where three jury members were seated behind a table. They were introduced as experts in nonverbal behavior and public speaking analysis. The participant was then instructed to prepare a 5-minutes speech in which he had to convince the jury that he was the perfect applicant for a vacant employment position. He was also informed that his presentation would be videotaped and audiotaped for later analysis. The participant was then taken back to room A, where the investigator left him alone to prepare his speech. After the 10 minutes preparation time, the investigator returned to the participant in room A and instructed him to go to the interview room B and to leave his notes behind. There he performed the public presentation (5 min) and the mental arithmetic task (5 min) while standing before the critical jury. The jury members were instructed to withhold any normal encouraging smiles and nods while the participant gave his presentation and to remain silent, with no leading questions when the participant paused. When the participant had finished the mental arithmetic task, the jury chairman told the participant to return to room A, where the exploration task was performed.

365 voices) through their headphones. We used both a daytime and a nighttime version of this VE. The nighttime environment is identical to the daytime environment except for the low light level (the sky was dark). Due to the use of a contemporary Counter-Strike game engine, the graphics are up to modern-day standards. The Counter-Strike game was modified, and all regular game features (opponents and strategic game elements like the gun, information on the number of bullets left, and the map) were removed. Also, most keyboard and mouse functions were disabled. The keys w, a, s, and d could be used, to move forward, backward, and sideward through the environment. The only remaining function of the mouse was to rotate the viewing direction around the vertical axis (to look around). The Counter-Strike game ran on a Dell Dimension XPS600 computer. The VEs were displayed on a 19-inch Iiyama Prolite E481S display. Sound was presented through Sennheiser EH 150 headphones. The VE exploration task The VE exploration task was performed in room A and lasted 10 minutes. The participant was instructed to imagine that he was a military scout, situated in an unknown village, whose task it was to perform a thorough reconnaissance of this village. This task served to ensure that the participants would perform a detailed visual inspection of the VE. The participant was unaware that his affective appraisal and recollection of scenic details would be tested at a later stage in the experiment. Room lighting remained on when the participant explored the daytime VE and was turned off when he explored the nighttime VE. After finishing the VE exploration task, the participant was asked to complete the semantic questionnaire and the yes–no recognition test. Next, a (second) saliva sample was obtained. This was done 35 minutes after the start of the experiment (20 minutes of stress=relaxation, 10 minutes of VE exploration, 5 minutes of filling out questionnaires), and 15 minutes after cessation of the stress=relaxation condition, when cortisol levels are known to peak.29 Finally, the participant was asked to complete the TSAI for the second time. Russell and Pratt semantic questionnaire

Control condition The protocol for the control groups was identical to the protocol for the experimental groups, except that the control groups did not perform the TSST, but browsed some magazines instead (for the purpose of relaxation), during a period of 20 minutes (the duration of the TSST). The virtual environment The VE used in this study was the Italy level of CounterStrike (Source Engine 7). This level represents a small village with a typical Italian architecture, with narrow streets, steep stairs, and with a marketplace in its center surrounded by houses (Fig. 1). Participants could walk through the streets of this village or enter and explore some of its houses. While exploring, the participants heard a simulation of the sound of their footsteps, as well as all sorts of background noise (music, a singing voice, a passing airplane, wind, murmuring

The affective appraisal of the VE was measured using a translated and shortened version of the semantic questionnaire developed by Russell and Pratt.15 This shortened version consists of 20 bipolar adjectives that are descriptive of the affective quality of physical environments. Each of the four basic affective appraisal states (arousing, sleepy, pleasant, unpleasant) is tested by five adjectives, using an 8-point semantic differential Likert scale. Recognition test The recognition test comprised a self-paced PowerPoint presentation of 20 screenshots taken from objects represented in the computer game Counter-Strike. Ten screenshots represented objects that participants actually encountered when they explored the virtual environment (Counter-Strike level Italy). The other 10 screenshots represented objects residing in levels of Counter-Strike that the participants had not

366

TOET ET AL.

FIG. 1. Screenshots of a village square (A, B) and an alley (C, D) in the daytime (A, C) and nighttime (B, D) virtual environments.

explored. The screenshots were randomly ordered within the presentation. The presentation was shown on the same monitor that was used for the exploration task to avoid any representation (color or contrast) differences. Participants were provided with a pencil and a score list. They could advance the presentation by pressing the space bar of a standard keyboard. They were instructed to mark on the score list for each object whether or not they had seen it in the VE. No feedback was provided to the participants as to the correctness of their answers. Similar recognition tests have been used in the literature to investigate the user experience of VEs.42,43 Participants The experimental sample consisted of 52 male volunteers, whose age ranged from 18 to 32 years (M ¼ 23.37, SD ¼ 3.27). A population consisting of young males was chosen because they are known to show the largest cortisol reactivity in response to stress and to rule out potential effects of menstrual

hormonal cycle.32 None of the participants had any experience with the computer game Counter-Strike. Several days before the experiment, all participants received written information and instructions about the experimental procedure. The information stated that the study involved the relationship between personal traits and the experience of a VE. It also stated that their task would be to navigate, explore, and evaluate a VE presented on a desktop computer, while some saliva samples would be taken and their heart rate would be measured. As a result, all participants were naı¨ve to the actual purpose and procedure of the experiment. The participants were instructed to refrain from eating and drinking coffee or tea for a period of at least 1.5 hours before the start of the experiment and not to eat any licorices or eggs during a period of at least 24 hours before the experiment, since these foods could compromise the saliva samples. Furthermore, participants were asked not to brush their teeth too vigorously, to prevent the infliction of open gum wounds, whereby blood could contaminate the saliva samples. The participants were urged to carefully follow these

IS A DARK VIRTUAL ENVIRONMENT SCARY?

367

FIG. 2. Normalized mean heart rate (beats per minute) as a function of elapsed time since the start of the experiment (time ¼ 0) for all four conditions. Heart rate was normalized by subtracting the mean heart rate determined for each individual during a period of 5 minutes before the start of the experiment (baseline). The Trier Social Stress Test was performed between 10 and 30 minutes after the start of the experiment. Stress (control) day (night) ¼ stress (control) group that navigated in the daytime (nighttime) VE.

instructions to ensure that the saliva samples would be suitable for further analysis. Participants were randomly assigned to one of four different conditions: two experimental (stress) conditions (the TSST followed by exploration of either the daytime or the nighttime VE) and two control (relaxation) conditions (a relaxation period followed by exploration of either the daytime or the nighttime VE). Each group contained 13 participants. We will refer to the groups that were subjected to the TSST as the stress groups and to groups in the relaxation condition as the control groups.

time. At the end of the experiment, the participants were extensively debriefed, paying special attention to participants in the stress condition. The nature of the study was explained in great detail, and the investigator made sure that the participants left the laboratory without any adverse effects.

Experimental procedure Experimental sessions lasted 1 hour. After arrival at the laboratory, the participant was taken to room A, where he first read and signed an informed consent form. Then, he put on the chest strap with the Polar Wearlink 31 wireless transmitter and the wrist strap with the Polar RS400 Running Computer. Next, a saliva sample was obtained by letting the participant gently chew on the Salivette dental cotton for 1.5 minutes. After that, he was asked to complete the TSAI,36 the general questionnaire, and the Game Experience questionnaire. Then the participant received a general instruction about the content of the two tasks he was asked to perform. Participants in the stress condition first performed the TSST, which lasted for 20 minutes. Participants in the control condition spent the same period reading magazines. The second task was the same for both groups and required the participant to explore a VE for a period of 10 minutes. Next, all participants performed the VE exploration task in room A. After finishing this task, the participants were asked to complete the semantic questionnaire and the yes–no recognition test. Next, a second saliva sample was obtained, and the participant was asked to complete the TSAI for the second

FIG. 3. The mean free salivary cortisol level (nmol=l) at the start of the experiment and 35 minutes after the start of the Trier Social Stress Test or control condition, per group. Stress (control) day (night) ¼ stress (control) group that navigated in the daytime (nighttime) VE.

368

TOET ET AL. Table 1. Mean Level of Free Salivary Cortisol (nmol=l), in the First and Second Sample, and Their Mean Absolute Difference, for Each Condition First sample

Stress day Stress night Control day Control night

Second sample

Absolute difference

M

SD

M

SD

M

SD

9.33 8.15 7.49 8.65

8.94 3.19 3.40 3.76

19.74 14.21 5.32 5.14

11.62 7.92 3.04 2.85

10.41 7.08 2.89 3.89

6.85 6.53 2.05 2.42

The experimental protocol was approved by TNO Human Factors internal review board on experiments with human participants and was in accordance with the Helsinki Declaration of 1975, as revised in 2000.44 The participants gave their informed consent prior to testing. The participants received a modest financial compensation for their participation. Results In this section, we first evaluate the results of the three different stress tests (heart rate, free salivary cortisol, the TSAI), to assess the extent to which the TSST evoked a stress response. Then we present the results of the affective appraisal and recognition tests. Stress induction Heart rate. To control for individual differences in physical fitness levels, we first assessed the mean heart rate for each individual during a period of 5 minutes before the start of the experiment. Heart rate was then normalized by subtracting this baseline (resting) value. Fig. 2 shows the normalized mean heart rate over the course of the experiment for all four experimental groups. The two stress groups showed an elevated heart rate over the period of the TSST stimulation (i.e., between 10 and 30 minutes after the start of the experiment). Shortly after cessation of the stressor, heart rates dropped to baseline. The two control (relaxation) groups showed a minor decrease in heart rate during the same period. This result indicates that the TSST procedure effectively provoked an increased heart rate. Salivary cortisol. Fig. 3 shows the mean free salivary cortisol level (nmol=l) at the start of the experiment and 5 minutes after finishing the VE exploration task (i.e., 35 minutes after the start and 15 minutes after the end of the TSST or

the control condition) for the two stress groups and the two control groups (the exact values and their standard deviations are given in Table 1). At the start of the experiment, there was no significant difference between the level of free salivary cortisol of the four groups. A repeated measures t test for mean cortisol level showed that the mean salivary cortisol levels significantly increased in response to the TSST simulation for both stress groups: stress day, t(12) ¼
5.48, p ¼ 0.00; stress night: t(12) ¼
2.89, p ¼ 0.01. The levels significantly decreased after the relaxation period for both control conditions: control day, t(12) ¼ 2.75, p ¼ 0.02; control night: t(12) ¼ 4.24, p ¼ 0.00. Thus, it appears that the TSST (stress condition) effectively provoked an increase of cortisol levels, whereas the relaxation task (control condition) effectively lowered cortisol levels. Translated State Anxiety Inventory. Table 2 shows the mean scores on the TSAI, both at the start (begin score) and at the end (end score) of the experiment, and the mean absolute difference between these two scores, for all four groups. All groups scored higher at the end of the experiment, which implies that their perceived anxiousness increased during the experiment. The difference between the begin and end scores was significant for the stress day and the control night groups, respectively, t(12) ¼
2.65, p ¼ 0.02; t(12) ¼
3.35, p ¼ 0.01, but not for the stress night and the control day groups. This means that the night condition evoked anxiety only in the control group and not in the stress group. Summarizing, the glucocorticoid (salivary cortisol level) and sympathetic (heart rate) components of stress response and the scores on the TSAI36 all indicate that the TSST procedure effectively elicited signs of distress. The TSAI indicates that the night condition evoked anxiety only in the control group, not in the stress group.

Table 2. Scores on the Translated State Anxiety Inventory at the Start and at the End of the Experiment, and Their Absolute Difference, for Each Condition Begin score

Stress day Stress night Control day Control night

End score

Absolute difference

M

SD

M

SD

M

SD

34.08 29.92 30.15 29.08

6.54 6.22 5.00 4.19

41.77 33.46 30.69 32.92

9.89 6.97 2.75 3.97

9.38 6.46 4.69 4.46

8.83 5.11 3.52 3.41

IS A DARK VIRTUAL ENVIRONMENT SCARY?

369

Table 3. Total Score on Each Scale of the Semantic Questionnaire, for Each Condition and All Factors Arousing

Stress day Stress night Control day Control night

Sleepy

Unpleasant

M

SD

M

SD

M

SD

M

SD

3.40 3.54 3.58 4.09

1.12 0.87 1.09 1.48

4.94 5.12 5.40 4.65

1.13 1.16 0.93 1.06

4.45 3.85 5.15 4.03

1.42 1.07 1.17 1.08

3.77 4.72 3.60 4.55

1.58 1.10 1.53 1.07

Affective appraisal. Table 3 lists the means and standard deviations of the total score on each scale of the semantic questionnaire for each condition and all factors. To assess the reliability of the semantic questionnaire,15,16 Cronbach’s a was calculated for each of its four scales. All were greater than 0.70, which is sufficiently high to preserve all items per scale. For each of the four scales, a 22 (stress leveltime-of-day level) ANOVA was performed. There was a significant difference among the four conditions in the mean total score on the scale pleasant, F(3, 48) ¼ 3.07, p ¼ 0.04, but not on the scales arousing, sleepy, and unpleasant. Participants in the stress groups appraised the VE as more sleepy and unpleasant, and as less arousing and pleasant, than did control participants. But these differences are not significant. For the scales arousing and sleepy, there were no significant stress or ambient lighting main effects and no significant interaction effects. There was a significant illumination effect for the scales pleasant, F(1, 48) ¼ 6.78, p ¼ 0.01, and unpleasant, F(1, 48) ¼ 6.59, p ¼ 0.01, but no significant stress or interaction effect. The only significant difference in the total score revealed by follow-up t tests was on the pleasant scale, between the control day and control night group, t(24) ¼ 2.54, p ¼ 0.02. This means that the nighttime VE was appraised significantly more unpleasant only by the control group, not by the stress group. Recognition Table 4 lists the mean number of correct responses for the yes–no recognition test and for each of the four experimental conditions. The number correct is the number of objects that participants correctly recognized as actually present in the part of the VE they explored, plus the number of objects they correctly dismissed as not being there. A 22 (stress levelambient lighting level) ANOVA and follow-up t tests Table 4. Mean Number Correct Responses on Yes–No Recognition Test and Standard Deviations for Each Experimental Condition Number correct

Stress day Stress night Control day Control night

Pleasant

M

SD

17.08 16.46 16.00 15.92

1.38 1.39 1.78 2.50

revealed no significant difference in the number correct responses between any of the four groups. Discussion In this study, we investigated the effects of ambient lighting and prior stress on the affective appraisal of a desktop VE. Participants were first exposed to either the TSST or to a nonstressful control task. Then they explored either a daytime or a nighttime version of a VE. Participants who were subjected to the TSST showed increased free salivary cortisol levels, higher heart beat rates, and higher scores on the TSAI, which are all signs of distress. Participants who were subjected to a relaxing control condition did not show any of these signs. The TSST thus appeared to evoke distress in the participants. Participants in the control group appraised the daytime VE as significantly more pleasant and the nighttime VE as significantly more unpleasant. However, there was no significant effect of time-of-day on the scales sleepy and arousing. This finding is in contrast with the results of Rohrman and Bishop,45 who found that people appraised a nighttime VE as more arousing than a daytime VE. While the appraisal of an environment as pleasant or unpleasant may be strongly determined by the lightness level or time of day, the appraisal as arousing or sleepy (the static quality) is probably less dependent on the time of day, since light and dark environments can both be static or dynamic. The results of the yes–no recognition test showed no significant difference in the number of correctly recognized details between any of the four different conditions. The reason for this lack of effect may be that the test details were too salient both in the daytime and in the nighttime environments, so that the task may have been too easy in all conditions. Differences in recognition scores may become apparent for a more difficult recognition task, possibly in combination with arousing events and emotional scenes.46 Summarizing, we find that the affective appraisal of a desktop VE was not significantly affected by the simulated nighttime lighting conditions, even after acute prior stress. The absence of an effect may be due to the combination of (a) an adaptation effect and (b) the lack of an engaging task. It has been observed that a prior affective experience biases the appraisal of the corresponding affective quality of an environment in a direction opposite to the value of the prior experience.47 The TSST was definitely an unpleasant experience for the participants in this study, which may have shifted their appraisal of the VE in the direction of pleasantness, thus obscuring a possible darkness effect.

370 Another explanation of the fact that the stress and control groups showed no difference in their affective appraisal of the VE may be that distressed participants do not focus on the environment in an affective sense due to a narrowing of attention. Maybe they need all their attentional resources to focus on the exploration task in the VE so that they appraise this environment in a neutral way. This issue may be resolved by assessing whether the immersion of the participants in a VE is indeed hampered by stress. The power of a game to elicit stress is largely dependent on the impact the game environment has on the player. If the player is not stimulated or motivated to put effort in the game and to perform well, the impact of stress manipulations will be limited. The intended positive and negative emotions will be evoked only when the player is immersed in the game or at least is committed to put effort in the task and attempt to complete the game. Maybe more differences in affective appraisal between the groups would have appeared if the stressor had been inherent in the VE itself (e.g., a virtual version of the TSST41) or had been directly related to the task.48 We conclude that merely lowering the ambient light level in a desktop VE is not sufficient to elicit anxiety. In this respect, desktop VE representations are different from immersive VE representations. This finding has implications for the use of desktop VE representations in serious gaming and training applications. Acknowledgments The authors thank Erik van der Spek for adapting the Counter-Strike level Italy (Source Engine 7) for the purpose of this experiment. They are also indebted to Roy Raymann and Ineke Klo¨pping-Ketelaars for their advice and help with the experimental design and the saliva sampling and analysis, to Etie¨nne Grasse`re and Susan Vrijkotte for their help with the heart rate registration equipment, and to Anja Langefeld for her support with the data analysis. This research has been supported by the GATE project, funded by the Netherlands Organization for Scientific Research (NWO) and the Netherlands ICT Research and Innovation Authority (ICT Regie).

TOET ET AL.

5. 6.

7. 8. 9.

10.

11.

12.

13.

14.

15.

16.

17.

18. 19.

Disclosure Statement No competing financial interests exist.

20.

References 1. Gratch J, Marsella S. Fight the way you train: the role and limits of emotions in training for combat. Brown Journal of World Affairs 2003; 10:63–75. 2. Wilfred LM, Hall R, Hilgers M, et al. (2004) Training in affectively intense virtual environments. In Richards G, eds. Proceedings of World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education 2004. Chesapeake, VA: Association for the Advancement of Computing in Education, pp. 2233–40. 3. Ulate SO. The impact of emotional arousal on learning in virtual environments. Dissertation, Naval Post Graduate School, Monterey, CA, 2002. 4. Cozens P, Neale R, Hillier D. Managing crime and the fear of crime at railway stations–a case study in South Wales

21.

22.

23.

(UK). International Journal of Transport Management 2003; 1:121–32. Gray JA. (1987) The psychology of fear and stress. Cambridge, UK: Cambridge University Press. Blo¨baum A, Hunecke M. Perceived danger in urban public space. The impact of physical features and personal factors. Environment & Behavior 2005; 37:465–86. Warr M. Dangerous situations: social context and fear of victimization. Social Forces 1990; 68:891–907. Nasar JL, Jones KM. Landscapes of fear and stress. Environment & Behavior 1997; 29:291–323. Grillon C, Pellowski M, Merinkangas K, et al. Darkness facilitates the acoustic startle reflex in humans. Biological Psychiatry 1997; 42:453–60. Grillon C, Merikangas K, Dierker L, et al. Startle potentiation by threat of aversive stimuli and darkness in adolescents: a multi-site study. International Journal of Psychophysiology 1999; 32:63–7. Mu¨hlberger A, Wieser MJ, Pauli P. Darkness-enhanced startle responses in ecologically valid environments: a virtual tunnel driving experiment. Biological Psychology 2007; 77:47–52. Niedenthal S. (2005) Shadowplay: simulated illumination in game worlds. Proceedings of DiGRA 2005 Conference: Changing Views—Worlds in Play. Vancouver, Canada: University of Vancouver Press. Knez I, Niedenthal PM. Lighting in digital game worlds: effects on affect and play performance. CyberPsychology & Behavior 2008; 11:129–37. Russell JA, Snodgrass J (1987) Emotion and the environment. In Stokols D, Altman I, eds. Handbook of environmental psychology. New York: Wiley, pp. 245–80. Russell JA, Pratt G. A description of the affective quality attributed to environments. Journal of Personality & Social Psychology 1980; 38:311–22. Russell JA, Ward LM, Pratt G. Affective quality attributed to environments: a factor analytic study. Environment & Behavior 1981; 13:259–88. Schwarz N, Clore GL. Mood, misattribution, and judgments of well-being: informative and directive functions of affective states. Journal of Personality & Social Psychology 1983; 45:513–23. Russell JA. Core affect and the psychological construction of emotion. Psychological Review 2003; 110:145–72. Core GL, Huntsinger JR. How emotions inform judgment and regulate thought. Trends in Cognitive Sciences 2007; 11:393–9. Schwarz N (1990) Feelings as information: informative and motivational functions of affective states. In Sorrentino RM, Higgins ET, eds. Handbook of motivation and cognition: foundations of social behavior. New York: Guilford Press, pp. 527–61. Lerner JS, Keltner D. Beyond valence: toward a model of emotion-specific influences on judgment and choice. Cognition & Emotion 2000; 14:473–93. Zillmann D. Excitation transfer in communication-mediated aggressive behavior. Journal of Experimental Social Psychology 1971; 7:419–34. van der Spek ED, Bos JE, van Emmerik ML, et al. (2007) Cybersickness affects the affective appraisal of a virtual environment. In So S, Cheung R, Chow E, et al. eds. Proceedings of the First International Symposium on Visually Induced Motion Sickness, Fatigue, and Photosensitive Epileptic Seizures (VIMS2007). Clear Water Bay, Kowloon, Hong Kong: Hong Kong University of Science & Technology, pp. 170–7.

IS A DARK VIRTUAL ENVIRONMENT SCARY? 24. Eysenck MW (2008) Anxiety: the cognitive perspective. Hove, UK: Psychology Press. 25. Mathews A, MacLeod C. Discrimination of threat cues without awareness in anxiety states. Journal of Abnormal Psychology 1986; 95:131–8. 26. Easterbrook JA. The effect of emotion on cue utilisation and the organisation of behaviour. Psychological Review 1959; 66:183–201. 27. Simoens VL, Istok E, Hyttinen S, et al. Psychosocial stress attenuates general sound processing and duration change detection. Psychophysiology 2007; 44:30–8. 28. Grillon C, Duncko R, Covington MF, et al. Acute stress potentiates anxiety in humans. Biological Psychiatry 2007; 62:1183–6. 29. Kirschbaum C, Pirke KM, Hellhammer DH. The Trier Social Stress Test: a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 1993; 28:76–81. 30. Williams A, Hagerty BM. Trier Social Stress Test: a method for use in nursing research. Nursing Research 2004; 53:277–80. 31. Holsboer F. The rationale for corticotropin-releasing hormone receptor (CRH-R) antagonists to treat depression and anxiety. Journal of Psychiatric Research 1999; 33:181–214. 32. Kudielka BM, Buske-Kirschbaum A, Hellhammer DH, et al. HPA axis responses to laboratory psychosocial stress in healthy elderly adults, younger adults, and children: impact of age and gender. Psychoneuroendocrinology 2004; 29:83–98. 33. Umeda T, Hiramatsu R, Iwaoka, T, al. Use of saliva for monitoring unbound free cortisol levels in serum. Clinica Chimica Acta 1981; 110:245–53. 34. Levine A, Zagoory-Sharon O, Feldman R, et al. Measuring cortisol in human psychobiological studies. Physiology & Behavior 2007; 90:43–53. 35. Dressendorfer RA, Kirschbaum C, Rohde W, et al. Synthesis of a cortisol-biotin conjugate and evaluation as a tracer in an immunoassay for salivary cortisol measurement. Journal of Steroid Biochemistry & Molecular Biology 1992; 43: 683–92. 36. van der Ploeg HM. (2000) Handleiding bij de Zelfbeoordelingsvragenlijst: een Nederlandse bewerking van de Spielberger State Trait Anxiety Inventory Manual to the Dutch version of the Spielberger State Trait Anxiety Inventory. Lisse, The Netherlands: Swets Test Publishers. 37. Spielberger CD. Assessment of state and trait anxiety: conceptual and methodological issues. Southern Psychologist 1985; 2:6–16.

371 38. Goldberg LR. The development of markers for the Big-Five factor structure. Psychological Assessment 1992; 4:26–42. 39. Swickert RJ, Rosentreter CJ, Hittner JB, et al. Extraversion, social support, and stress. Personality & Individual Differences 2002; 32:877–91. 40. Ode S, Robinson MD. Agreeableness and the self-regulation of negative affect: findings involving the neuroticism=sosomatic distress relationship. Personality & Individual Differences 2007; 43:2137–48. 41. Kelly O, Matheson K, Martinez A, et al. Psychosocial stress evoked by a virtual audience: relation to neuroendocrine activity. CyberPsychology & Behavior 2007; 10:655–62. 42. McCreary FA, Williges RC (1998) Effects of age and field-ofview on spatial learning in an immersive virtual environment. Proceedings of the Human Factors and Ergonomics Society 42nd Annual Meeting. Santa Monica, CA: Human Factors & Ergonomics Society, pp. 1491–5. 43. Tortell R, Luigi DP, Dozois A, et al. The effects of scent and game play experience on memory of a virtual environment. Virtual Reality 2007; 11:61–8. 44. World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. Journal of the American Medical Association 2000; 284:3043–5. 45. Rohrmann B, Bishop ID. Subjective responses to computer simulations of urban environments. Journal of Environmental Psychology 2002; 22:319–31. 46. Lupien SJ, Maheu F, Fiocco A, et al. The effects of stress and stress hormones on human cognition: implications for the field of brain and cognition. Brain & Cognition 2007; 65: 209–37. 47. Russell JA, Lanius UF. Adaptation level and the affective appraisal of environments. Journal of Environmental Psychology 1984; 4:119–35. 48. Morris CS, Hancock PA, Shirkey EC. Motivational effects of adding context relevant stress in PC-based game training. Military Psychology 2004; 16:135–47.

Address correspondence to: Dr. Alexander Toet Department of Applied Perception TNO Human Factors Kampweg 5 Soesterberg The Netherlands 3769DE E-mail: [email protected]